Process for self-aligned manufacture of integrated electronic devices

Abstract

 A process for self-aligned manufacturing of integrated electronic devices includes: forming, in a semiconductor wafer (20) having a substrate (21), insulation structures (27) that delimit active areas (25,26) and project from the substrate (21); deposing a first conductive layer (39), which coats the insulation structures (27) and the active areas (25,26); and partially removing the first conductive layer (28). In addition, in the insulation structures (27) recesses (32,33) are dug before deposing said first conductive layer (28).

Description

[0001] The present invention relates to a process for self-aligned manufacture of integrated electronic devices.

[0002] As is known, in modern microelectronics, reduction of the overall dimensions of devices is one of the main objectives. In particular, in the fabrication of memories of a non-volatile type, it is important to minimize the overall dimensions of each memory cell. The need to obtain an increasingly wider integration scale entails, however, certain difficulties. In some cases, for example, the alignment of the masks used in the different processing steps using traditional processes calls for a precision that, in practice, is frequently not possible to achieve. In particular, a major problem is to align the masks normally utilized, on the one hand, for defining the active areas accommodating the memory cells and, on the other, for shaping the polysilicon layer extending on top of the active areas and forming the floating gates of the cells.

[0003] So-called self-aligned processes have consequently been developed, and enable the more critical masking steps to be eliminated, exploiting the surface conformation of the wafer. For greater clarity, reference may be made to Figures 1 to 4, showing a semiconductor wafer 1 having a substrate 10, for example of monocrystalline silicon. The wafer 1 comprises conductive active areas 2, insulated by shallow-trench-insulation (STI) structures 3, or else, alternatively, by insulation structures formed through local oxidation of silicon (LOCOS). In practice, the insulation structures 3 comprise trenches of a preset depth, filled with silicon dioxide. In either case, the insulation structures 3 project from the surface 4 of the wafer 1, adjacent to the active areas 2; in this way, the insulation structures 3 define recesses 5 exactly on top of the active areas 2.

[0004] Channel regions of memory cells (not illustrated herein) are made inside the active areas 2 by implanting and diffusing dopant species and using thermal oxidation; then a thermal oxidation provides a gate oxide layer 7, of the thickness of a few nanometers. Subsequently, a conductive polysilicon layer 8 is deposited, as illustrated in Figure 2.

[0005] The conductive layer 8 fills the recesses 5 and has a thickness such as to cover completely the projecting portions of the insulation structures 3.

[0006] Next (Figure 3), a chemical-mechanical-polishing (CMP) planarization is performed, which is stopped when the insulation structures 3 are again exposed. In this way, the entire polysilicon layer 8 is removed, except for residual portions, which occupy the recesses 5 and are consequently perfectly aligned to the active areas 2. In practice, the residual portions of the polysilicon layer 8, which are insulated from the respective active areas 2 thanks to the oxide layer 7, form floating gates 11 of the memory cells.

[0007] The process further comprises forming an insulating layer 12, which coats the floating gates 11 of the polysilicon layer 8, and depositing a further polysilicon layer, which is in turn defined for forming control gates 13 of the memory cells.

[0008] The known self-aligned processes, although advantageous as regards the possibility of increasing the integration scale, present, however, other limitations. Traditional processes, in fact, enable passive components (normally resistors and capacitors) to be formed on top of the insulating structures. In particular, these components and floating gates of the memory cells may be formed starting from the same polysilicon layer using a single mask. This is particularly useful for forming parts of read/write circuits of the memory cells, which are normally integrated in the same wafer, but must withstand much higher voltages and currents. The gate oxide is in fact too thin for eliminating the inevitable capacitive coupling of the high-voltage passive elements with the substrate and is highly subject to breakdown if subjected to high voltages. In addition, traditional processes enable standard cells and high-performance cells to be formed in the same wafer. In particular, in the high-performance cells, the floating terminal is shaped so as to extend in part also outside of the active areas and is consequently better coupled to the control gate: these cells may consequently be driven more rapidly and/or with lower voltages.

[0009] It is, however, evident that known self-aligned processes do not enable either passive components or high-performance cells to be formed on top of the insulating structures. On the one hand, in fact, the CMP treatment removes completely the polysilicon overlying the insulating structures, where no conductive material is available to form electrical components; it is consequently necessary to depose and define a new polysilicon layer. On the other hand, precisely because the processes are self-aligned, the recesses where the floating gates of the cells are formed have the same dimensions as the underlying active areas and consequently it is not possible to improve the coupling.

[0010] The aim of the present invention is to provide a self-aligned process for manufacturing integrated electronic devices which is free from the drawbacks described above.

[0011] According to the present invention, there is provided a self-aligned process for manufacturing integrated electronic devices and an integrated electronic device, as defined in claims 1 and 19, respectively.

[0012] For a better understanding of the invention, some embodiments thereof are now described purely by way of non-limiting example and with reference to the attached drawings, wherein:

• Figures 1 to 16 are cross-sectional views through a semiconductor wafer in successive fabrication steps of a process according to a first embodiment of the present invention; and
• Figures 17 and 18 are cross-sectional views through a semiconductor wafer in successive fabrication steps of a process according to a different embodiment of the present invention.

[0013] Hereinafter, the process according to the invention is used to manufacture non-volatile memories, in particular of EEPROM or flash type; this is not, however, to be considered limiting, in so far as the process may be used also in electronic devices of another type.

[0014] With reference to Figures 5 to 16, a semiconductor wafer 20, preferably of silicon, has a substrate 21, for example of P type. Initially, a hard mask 22 is formed on the wafer 20; the hard mask 22 comprises a pad oxide layer 22a and a silicon nitride layer 22b and has openings 23. Using the hard mask 22, the substrate 21 of the wafer 20 is etched, and trenches 24 are opened, which delimit memory active areas 25 and circuitry active areas 26, where memory cells and, respectively, read/write circuits and control circuits (Figure 6) will subsequently be formed.

[0015] After a thermal-oxidation step, optimizing the profile of the trenches 24, the trenches 24 are filled with dielectric material, here silicon dioxide. The wafer 20 is then planarized with a first chemical-mechanical-polishing (CMP) treatment; in particular, the CMP treatment is interrupted when the hard mask 22 has been reached, as illustrated in Figure 7. At this point, in practice, the active array areas 25 and circuitry-active areas 26 are delimited laterally by trench insulation structures 27, which extend in part inside the substrate 21 and have projecting portions 27a projecting at the top from the substrate 21 and aligned with the hard mask 22.

[0016] Subsequently, a resist mask 28 is formed on top of the wafer 20 and has first and second openings 30, 31 (see Figure 8). In detail, the first openings 30 are formed on top of some of the memory active areas 25, where high-performance cells are to be formed. More precisely, the first openings 30 expose portions of the hard mask 22 that overlie these active memory areas 25 and, furthermore, laterally expose the projecting portions 27a of the insulation structures 27 adjacent to them. The second openings 31, instead, centrally expose the projecting portions 27a of the insulation structures 27, which delimit the circuitry active areas 26. The remaining memory active areas 25, designed to accommodate standard cells, are covered by the resist mask 28.

[0017] Then the exposed silicon-dioxide portions are etched in a controlled way, as illustrated in Figure 9. In this step, in particular, first and second recesses 32, 33 are dug inside the insulation structures 27, which delimit the memory active areas 25 and, respectively, the circuitry active areas 26. In practice, the first recesses 32 are delimited at the bottom and on one side by the respective insulation structures 27 and, on the opposite side, by portions of the hard mask 22, which coat memory active areas 25. The second recesses 33 are, instead, dug completely inside the insulation structures 27, defining the circuitry active areas 26. In greater detail, the second recesses 33 are opened and accessible at the top and are delimited laterally and at the bottom by the respective insulation structures 27.

[0018] Next, the resist mask 28 and the hard mask 22 are removed, as illustrated in Figure 10. At this point, in practice, the first recesses 32 are connected to one another and form, in pairs, cavities 34 above the respective memory active areas 25; furthermore, third recesses 35 are defined on top of the memory active areas 25 intended to accommodate standard memory cells, and are delimited laterally by pairs of insulation structures 27.

[0019] In a known way, ion-implantation and diffusion are then performed for forming channel regions of memory cells (not illustrated herein for convenience) in the memory active areas 25; simultaneously, electronic components are provided in the circuitry-active areas 26 and form read/write and control circuits 36, here indicated only schematically.

[0020] Subsequently, a gate oxide layer 37 with a thickness of a few nanometers is grown thermally, and coats both the memory active areas 25 and the circuitry active areas 26 (Figure 11). A first polysilicon layer 39 is then deposited on the wafer 20, coats the entire wafer 20 and fills the second and third recesses 33, 35 and the cavities 34.

[0021] The wafer 20 is then planarized with a second CMP treatment, which is stopped when the insulation structures 27 are again exposed, as illustrated in Figure 12. In this step, the first polysilicon layer 39 is removed completely, except for residual portions inside the second recesses 33, the cavities 34 and the third recesses 35, which form, in the first case, resistors 40 and first plates 41a of capacitors, and in the other cases, floating gates 44a, 45a of high-performance memory cells and standard memory cells, respectively. In this way, in practice, just one deposition of polysilicon, followed by a planarization step, enables conductive regions to be formed which extend entirely (resistors 40 and first plates 41a) or partially (floating gates 44a) on top of insulation structures 27. The steps described above are moreover self-aligned, in so far as they are formed by exploiting the surface conformation of the wafer 20.

[0022] Then a dielectric layer 47 and a second polysilicon layer 48 are deposited (Figure 13) and selectively etched for forming capacitors 41, high-performance cells 44, and standard cells 45. In particular, Figure 14, starting from the second polysilicon layer 48, second plates 41b are formed on top of the first plates 41a, and control gates 44b, 45b are formed on top of the floating gates 44a, 45a of high-performance cells 44 and standard cells 45, respectively. In addition, the second plates 41b and the control gates 44b, 45b are insulated from the underlying conductive regions (first plates 41a, floating gates 44a, 45a) by respective residual portions 47' of the dielectric layer 47. Clearly, the floating gates 44a and control gates 44b of the high-performance cells have a greater capacitive coupling than those of the standard cells, since they have a larger surface. They extend, in fact, beyond the respective active areas 25 and occupy the first recesses 32 of the adjacent insulation structures 27.

[0023] The process then comprises depositing a protective dielectric layer 50, for example of silicon dioxide, and opening contacts 51 through the protective layer 50, figure 15. Finally, the wafer 20 is divided into individual dice 52, as illustrated in Figure 16; each die 52 comprises a respective electronic device, which, in the described embodiment, is a non-volatile memory.

[0024] The process according to the invention is clearly advantageous, because, through the addition of just one masked etch of the insulation structures 27, it enables both passive components presenting excellent insulation from the substrate 21 and memory cells with differentiated characteristics and performance to be formed on the same wafer 20.

[0025] In particular, the passive components (resistors 40 and capacitors 41) may operate with high voltages, without any risk of breakdown of the insulating dielectric and, furthermore, with a substantially negligible capacitive coupling to the substrate 21. These components are consequently suited for being used in read/write circuits, for example for forming charge pumps. As regards, instead, the memory cells, the process enables cells with high capacitive coupling between the control gate and the floating gate to be formed, in addition to the standard cells.

[0026] In this case, the high capacitive coupling is useful because the memory cells formed in this way may be driven with low voltages and hence have optimized performance. Memory cells of this type are particularly advantageous in the case of so-called "embedded" memories, which also integrate high-complexity logic circuits, such as, for example, microcontrollers or digital signal processors (DSPs).

[0027] In addition, the definition of the resist mask 28 for etching the insulation structures 27 is not critical and does not present problems of alignment with the active areas. Finally, the process is self-aligned and consequently enables standard cells of extremely contained dimensions to be formed.

[0028] Figures 17 and 18, where parts equal to those already illustrated are designated by the same reference numbers, show a different embodiment of the process according to the invention. In this case, after the insulation structures 27 have been formed and the wafer 20 has been planarized via the first CMP treatment, as already described previously, a first resist mask 55 is deposited and defined, to expose only a part of insulation structures 27 delimiting memory active areas 25; the insulation structures which delimit the circuitry-active areas 26 are instead protected (Figure 17). By a first controlled etch, the first recesses 32 are then formed.

[0029] After the first resist mask 55 and the hard mask 22 have been removed, a second resist mask 56 is formed on top of the wafer 20. Now, all the memory active areas 25 and the respective insulation structures 27 are protected, while central portions of the insulation structures, which delimit the circuitry-active areas 26, are left exposed. The wafer 20 is again etched in a controlled way, and the second recesses 33 are dug. The second resist mask 56 is then removed, and the procedure ends with deposition of the first polysilicon layer 28, second CMP treatment, and formation of passive components and memory cells, as already described with reference to Figures 13 to 16.

[0030] Thereby, the process enables recesses having a differentiated depth to be formed. In particular, it is possible to control the first etch, which is more critical, with greater precision. The first recesses 32, in fact, must accommodate a polysilicon layer of thickness sufficient for withstanding the CMP treatment without undergoing damage, but at the same time must never have a depth such as to uncover the memory active areas 25. According to the design specifications of the individual storage device, instead, it could be convenient to form second deeper recesses 33.

[0031] Finally, it is evident that modifications and variations may be formed to the process described, without thereby departing from the scope of the present invention. In particular, the steps of masked etching for opening the first and second recesses 32, 33 may be performed either before or after removal of the hard mask 22. The first recesses 32 could be dug also only in the insulation structures 27 which delimit one side of the memory-active areas 25, and not in those which delimit the other sides; in practice, for each memory-active area, only a first recess 32 is defined. In addition, the storage devices obtained according to the present process need not necessarily comprise both passive components formed on top of the insulation structures and high-performance memory cells; instead, the process may also be exploited for forming only resistors, only capacitors, or else only high-performance memory cells. Finally, as already mentioned previously, the process may be used also for forming devices other than non-volatile memories, such as for example volatile memories.

Claims

1. A process for self-aligned manufacturing integrated electronic devices, comprising the steps of:

- forming, in a semiconductor wafer (20) having a substrate (21), insulation structures (27) delimiting active areas (25, 26) and projecting from said substrate (21);

- forming a first conductive layer (39) coating said insulation structures (27) and said active areas (25, 26); and

- partially removing said first conductive layer (28);

characterized by, prior to forming said first conductive layer (28), forming recesses in said insulation structures (27).

2. The process according to claim 1, characterized in that said step of forming said first conductive layer (39) comprises depositing said first conductive layer and filling said insulation structures (27).

3. The process according to claim 1 or 2, characterized in that said step of forming recesses comprises removing side portions (27a) of said insulation structures (27) so as to form first recesses (32).

4. The process according to claim 3, characterized in that said first recesses are defined on top and at the side of respective said active areas (25).

5. The process according to claim 3 or 4, characterized in that said step of partial removing comprises forming first conductive regions (44a), which extend inside at least one of said first recesses (32) and on top of a respective said active area (25).

6. The process according to claim 5, characterized in that said first conductive regions (44a) are floating gates of respective first memory cells (44).

7. The process according to any one of the preceding claims, characterized in that said step of forming recesses comprises removing central portions (27b) of said insulation structures (27) so as to form second recesses (33).

8. The process according to claim 7, characterized in that said second recesses (33) are delimited laterally and at the bottom by respective said insulation structures (27).

9. The process according to claim 8 or 9, characterized in that said step of partial removing comprises forming second conductive regions (40, 41a), accommodated inside respective said second recesses (33).

10. The process according to claim 9, characterized in that said second conductive regions (40) comprise resistors.

11. The process according to claim 9 or 10, characterized in that said second conductive regions (41a) comprise first plates of respective capacitors (41).

12. The process according to any one of the preceding claims, characterized in that said step of forming recesses comprises performing at least one first masked etch of said insulation structures (27).

13. The process according to claim 12, characterized in that said step of digging comprises performing a second masked etch of said insulation structures (27).

14. The process according to any one of the preceding claims, characterized in that said step of partially removing comprises planarizing said wafer (20).

15. The process according to any one of the preceding claims, characterized in that said step of partially removing is followed by the steps of:

- forming a dielectric layer (47) on top of said wafer (20);

- forming a second conductive layer (48) on top of said dielectric layer (47); and

- selectively removing said dielectric layer (47) and said second conductive layer (48).

16. The process according to claim 15, characterized in that said step of selectively removing comprises forming third conductive regions (41b, 44b).

17. The process according to claims 6 and 16, characterized in that said third conductive regions (44b) comprise control gates of said memory cells (44).

18. The process according to claims 11 and 16, characterized in that said third conductive regions (41b) comprise second plates of said capacitors (41).

19. An integrated electronic device comprising:

- a semiconductor body (20) having a substrate (21); and

- a plurality of insulation structures (27) delimiting active areas (25, 26) and having respective portions (27a) projecting from said substrate (21); characterized in that said insulation structures (27) have respective recesses (32, 33), which accommodate at least partially conductive regions (40, 41a, 44a).

20. The device according to claim 19, characterized in that said recesses (32) are defined laterally with respect to respective projecting portions (27a) of said insulation structures (27).

21. The device according to claim 19 or 20, characterized in that said conductive regions (44a) comprise terminals of memory cells (44) arranged on top of respective said active areas (25) and extending laterally inside at least one of said recesses (32).

22. The device according to claim 19, characterized in that said recesses (33) are defined centrally with respect to respective said projecting portions (27a) of said insulation structures (27).

23. The device according to claim 22, characterized in that said conductive regions (40, 41a) are entirely accommodated inside respective said recesses (33).

24. The device according to claim 22 or claim 23, characterized in that said conductive regions (40) comprise resistors.

25. The device according to any one of claims 22 to 24, characterized in that said conductive regions (41a) comprise first plates of capacitors (41).

Drawing